Yury Kartynnik
Staff Software Engineer at Google
Zurich, Zurich, Switzerland
Summary
👤
Senior🎓
Top SchoolYury Kartynnik is a Staff Software Engineer based in Zurich with 11 years of experience building on-device and infrastructure-level ML systems across vision, NLP, and audio. He brings deep academic roots (PhD in mathematics) to pragmatic engineering, having shipped real-time mobile segmentation at AIMatter and led on-device ML and safety work at Google, including contributions to Google's AI Red Team and the Project Big Sleep initiative. Yury is an active low-level ML engineer in open source—his XNNPACK commits add optimized kernels and expose transpose/quantized deconvolution ops, reflecting expertise in performance-sensitive inference. He combines strong theoretical background in graph theory and computational complexity with hands-on C++/Python production experience, and is known for turning algorithmic insight into efficient, testable kernels for resource-constrained devices.
11 years of coding experience
14 years of employment as a software developer
Doctor of Philosophy - PhD Mathematics, Doctor of Philosophy - PhD Mathematics at Belarusian State Universitya la Masters Mathematics and Computer Science, a la Masters Mathematics and Computer Science at Yandex School of Data AnalysisHigh School Physics, High School Physics at Minsk 51 Secondary School
Russian, Russian, English, German, Ukrainian
Github Skills (8)
neural-network10
cpu10
inference10
cprogramming-language9
c-language9
unit-testing8
unit-test8
simd7
Programming languages (16)
Github contributions (5)
google/XNNPACK
Nov 2020 - Nov 2021
High-efficiency floating-point neural network inference operators for mobile, server, and Web
Role in this project:
ML Engineer
ML EngineerContributions:7 commits in 1 year
Contributions summary:Yury's commits focus on implementing and testing optimized kernels for the `xnnpack` library, which targets high-efficiency neural network inference. The primary contribution is the development of a basic scalar implementation for the CHW-to-HWC Depth-to-Space transformation, along with corresponding testing infrastructure. Furthermore, the commits expose XNNPACK transpose convolution implementation as a `TRANSPOSE_CONV` builtin op and expose quantized deconvolution via the subgraph API. This indicates a focus on low-level optimization and enabling new operators within the XNNPACK ecosystem.
multithreadingsimdtensorflowcpufloating
kartynnik/space2super
Nov 2014 - Jun 2022
Contributions:20 commits, 7 pushes in 7 years 7 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.